{"assertions":[{"assertion_index":0,"assertion_text":"Apple's research group has now moved that ceiling a long way: it ran Meta's Muse Spark Open, an 816-billion-parameter model (the size of a lab's top tier), across four Mac Studios with M5 Ultra chips and got more than 36 tokens per second, the word fragments models emit Apple Machine Learning Research.","citation_check":null,"claim_ids":[80097,80098,80099],"link_only_urls":[]},{"assertion_index":1,"assertion_text":"Muse Spark Open is a mixture-of-experts model, meaning only about 42 billion of its 816 billion parameters do work on any given token, so the machines must hold everything in memory but compute only a slice per word, and the M5 Ultra Mac Studio can be configured with up to 512 GB of memory, the kind of room a four-way split of a model this size needs Hugging Face Blog Nate Jones.","citation_check":null,"claim_ids":[57658,57661,57662,79485,79486,79487,79488],"link_only_urls":[]},{"assertion_index":2,"assertion_text":"Apple's researchers tied the four boxes together with MLX (Apple's software for running AI models on its own chips) and RDMA (a way for one machine to read another's memory directly), a capability the exo project says Apple promotes on its product pages as low-latency RDMA over Thunderbolt 5 Apple Machine Learning Research Latent Space.","citation_check":null,"claim_ids":[52403,80097,80098,80099],"link_only_urls":[]},{"assertion_index":3,"assertion_text":"The hardware then turns that fit into usable speed: Apple attributes an up to 4.6 times shorter wait before the first word to new accelerator units on the graphics chip, and 1.6 times faster generation to memory bandwidth, both measured against the older M3 Ultra Apple Machine Learning Research.","citation_check":null,"claim_ids":[80097,80098,80099],"link_only_urls":[]},{"assertion_index":4,"assertion_text":"Priced against hosted Muse Spark, the closed sibling, at $4.25 per million output tokens, that would displace roughly $13 a day in output fees, and at Meta's $0.20 contributor tier the figure drops well under a dollar Big Technology Simon Willison's Weblog.","citation_check":null,"claim_ids":[42896,67854],"link_only_urls":[]},{"assertion_index":5,"assertion_text":"The supplied pricing evidence gives no hosted price for Muse Spark Open itself, even though the same open weights can now be rented through Amazon SageMaker JumpStart, and nothing here prices the four-Mac cluster, so any saving has to be set against your own hardware quote AWS Machine Learning Blog.","citation_check":null,"claim_ids":[79703,79704],"link_only_urls":[]},{"assertion_index":6,"assertion_text":"That rental option actually sharpens the real case, which is control: a hosted deployment, SageMaker included, still sends your documents off site, while a cluster you own means no rate limits, no surprise repricing by a vendor, and data that never leaves the building, which is why confidentiality-bound professional services firms are already buying Mac Minis to run local models Nate Jones.","citation_check":null,"claim_ids":[15124],"link_only_urls":[]},{"assertion_index":7,"assertion_text":"Reflection AI launched Beam on October 5 as a US-trained open-weight (downloadable model weights) answer to Chinese models, and this week the plumbing caught up: Hugging Face published its welcome post and shipped native Beam support in Transformers 5.20.1, a patch that also repairs a 5.20.0 break in loading models on older versions of PyTorch, the underlying math library TechCrunch AI Hugging Face Blog Transformers Releases.","citation_check":null,"claim_ids":[73152,73153,73154,73155,73158,80333,80334,80335,80388,80391,80392],"link_only_urls":[]},{"assertion_index":8,"assertion_text":"Beam fires 23 billion of its 501 billion parameters per token, and its 1 million token context (how much text the model can read at once) comes from a mix of layers, 43 that look only at nearby text and nine that scan the whole document Hugging Face Blog.","citation_check":null,"claim_ids":[80333,80334,80335],"link_only_urls":[]},{"assertion_index":9,"assertion_text":"The score picture is disputed: Reflection says Beam scores as well as the best Chinese open models on hard reasoning tests while costing less to run, while Import AI's read is that it trails the Chinese model GLM-5.3 on benchmarks and leans on a token-efficiency claim, meaning it reaches the same answer while generating fewer word fragments TechCrunch AI Import AI.","citation_check":null,"claim_ids":[73152,73153,73154,73155,73158,76772],"link_only_urls":[]},{"assertion_index":10,"assertion_text":"Google Research has built VERID, a digital twin (a live software copy of the physical lab) on the open-source PyLabRobot, so an AI agent can dry-run and safety-check a procedure before a robot touches a pipette, and in aptamer experiments (short DNA strands selected to bind a target) confirmed hit rates, the share of designed candidates that actually bound the target, reached 33.5% and 42.1% after the agents spotted and corrected their own wet-lab failures Google Research Blog.","citation_check":null,"claim_ids":[80512,80513,80514],"link_only_urls":[]},{"assertion_index":11,"assertion_text":"An earlier AI co-scientist handed experts drug targets that held up at the bench; the new turn is the agent running the bench itself AI Engineer.","citation_check":null,"claim_ids":[16112],"link_only_urls":[]},{"assertion_index":12,"assertion_text":"- Independent long-context results for Reflection's Beam would test the token-efficiency claim against the disputed benchmark picture. Import AI","citation_check":null,"claim_ids":[76772],"link_only_urls":[]},{"assertion_index":13,"assertion_text":"- Google Research has promised closed-loop discovery of tight-binding proteins; a published binder with a named target is the proof point. Google Research Blog","citation_check":null,"claim_ids":[80512,80513,80514],"link_only_urls":[]},{"assertion_index":14,"assertion_text":"- The supplied pricing evidence gives no hosted price for Muse Spark Open; a listing at or below Muse Spark's $4.25 output rate, on SageMaker or elsewhere, would shrink the local cost case further. Big Technology AWS Machine Learning Blog","citation_check":null,"claim_ids":[67854,79703,79704],"link_only_urls":[]},{"assertion_index":15,"assertion_text":"- Black Forest Labs has now shipped FLUX 3 Image Klein, the 5.6B open-weight model, after promising an open-weight variant at the October 1 launch; independent tests of its 4K generation and editing claims are the next check. Latent Space Black Forest Labs","citation_check":null,"claim_ids":[73418,73420,80117,80118,80119],"link_only_urls":[]},{"assertion_index":16,"assertion_text":"- Hugging Face's follow-on Transformers patches should show whether the older-PyTorch loading break in 5.20.0 was fully closed by 5.20.1. Transformers Releases","citation_check":null,"claim_ids":[80388,80391,80392],"link_only_urls":[]},{"assertion_index":17,"assertion_text":"Apple's 36 tokens per second is a vendor-reported figure, and the figures available here do not say whether it covers one conversation or many served at once, which changes the cost arithmetic above Apple Machine Learning Research.","citation_check":null,"claim_ids":[80097,80098,80099],"link_only_urls":[]},{"assertion_index":18,"assertion_text":"The 512 GB memory figure is the top M5 Ultra configuration, not a stated spec of Apple's test machines Nate Jones.","citation_check":null,"claim_ids":[57658,57661,57662],"link_only_urls":[]},{"assertion_index":19,"assertion_text":"The $4.25 and $0.20 output prices are for hosted Muse Spark tiers, not the Open variant, and the daily-saving figures are hypotheticals built on them with no cluster price to set against; input fees ($1.25 per million on hosted Muse Spark) are left out because the evidence gives no figure for how fast the cluster processes input Big Technology Simon Willison's Weblog.","citation_check":null,"claim_ids":[42896,67854],"link_only_urls":[]},{"assertion_index":20,"assertion_text":"Google's hit rates come from its own writeup, the hit rate before the agents' corrections is not in the evidence here, and the binder goal is a forecast Google Research Blog.","citation_check":null,"claim_ids":[80512,80513,80514],"link_only_urls":[]}],"brief_date":"2026-10-17","bundle_id":"evidence-bundle-v2-767532bd1d29c83c379a3240f6ddd1cd5e7b1ad93834070620d3296a6ab61220","claims":[{"assertion_indexes":[0,2,3,17],"citation_authority_claim_id":80097,"citation_checks":[],"claim_id":80097,"claim_text":"Apple Research reports that running the Muse Spark Open model (816B parameters) on a cluster of four Mac Studios with M5 Ultra chips generates over 36 tokens per second.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:35:37.132837+00:00","source_name":"Apple Machine Learning Research","source_url":"https://machinelearning.apple.com/research/mlx-distributed-inference"},{"assertion_indexes":[0,2,3,17],"citation_authority_claim_id":80098,"citation_checks":[],"claim_id":80098,"claim_text":"Apple Research states that the M5 Ultra Mac Studio achieves up to 4.6 times faster time-to-first-token compared to the M3 Ultra Mac Studio due to GPU Neural Accelerators.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:35:37.132837+00:00","source_name":"Apple Machine Learning Research","source_url":"https://machinelearning.apple.com/research/mlx-distributed-inference"},{"assertion_indexes":[0,2,3,17],"citation_authority_claim_id":80099,"citation_checks":[],"claim_id":80099,"claim_text":"Apple Research states that the M5 Ultra Mac Studio provides 1.6 times faster generation speed (tokens per second) than the M3 Ultra Mac Studio due to increased unified memory bandwidth.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:35:37.132837+00:00","source_name":"Apple Machine Learning Research","source_url":"https://machinelearning.apple.com/research/mlx-distributed-inference"},{"assertion_indexes":[1,18],"citation_authority_claim_id":57658,"citation_checks":[],"claim_id":57658,"claim_text":"The base Mac mini is equipped with the new M6 chip, while more powerful models like the Mac mini, Mac Studio, and Mac Pro variants use the M5, M5 Pro, M5 Max, and M5 Ultra chips.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-08T06:26:50.208250+00:00","source_name":"Nate Jones","source_url":"https://www.youtube.com/watch?v=1lO8aNSLPJc"},{"assertion_indexes":[1,18],"citation_authority_claim_id":57661,"citation_checks":[],"claim_id":57661,"claim_text":"The M6 Mac mini offers 16 to 32 GB of unified memory, while the M5 Pro Mac mini reaches 64 GB with 307 GB/s of memory bandwidth.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-08T06:26:50.208250+00:00","source_name":"Nate Jones","source_url":"https://www.youtube.com/watch?v=1lO8aNSLPJc"},{"assertion_indexes":[1,18],"citation_authority_claim_id":57662,"citation_checks":[],"claim_id":57662,"claim_text":"The Mac Studio with M5 Max reaches 128 GB of memory, and the M5 Ultra configuration reaches 512 GB with 1.2 TB/s of bandwidth.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-08T06:26:50.208250+00:00","source_name":"Nate Jones","source_url":"https://www.youtube.com/watch?v=1lO8aNSLPJc"},{"assertion_indexes":[1],"citation_authority_claim_id":79485,"citation_checks":[],"claim_id":79485,"claim_text":"Muse Spark Open has a total parameter size of approximately 816 billion.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-16T06:48:14.640621+00:00","source_name":"Hugging Face Blog","source_url":"https://huggingface.co/blog/meta-muse-spark-open"},{"assertion_indexes":[1],"citation_authority_claim_id":79486,"citation_checks":[],"claim_id":79486,"claim_text":"Muse Spark Open activates approximately 42 billion parameters per token.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-16T06:48:14.640621+00:00","source_name":"Hugging Face Blog","source_url":"https://huggingface.co/blog/meta-muse-spark-open"},{"assertion_indexes":[1],"citation_authority_claim_id":79487,"citation_checks":[],"claim_id":79487,"claim_text":"Each layer in Muse Spark Open contains 256 experts, with a router selecting 8 experts for each token.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-16T06:48:14.640621+00:00","source_name":"Hugging Face Blog","source_url":"https://huggingface.co/blog/meta-muse-spark-open"},{"assertion_indexes":[1],"citation_authority_claim_id":79488,"citation_checks":[],"claim_id":79488,"claim_text":"Muse Spark Open supports a context window of up to 1 million tokens.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-16T06:48:14.640621+00:00","source_name":"Hugging Face Blog","source_url":"https://huggingface.co/blog/meta-muse-spark-open"},{"assertion_indexes":[2],"citation_authority_claim_id":52403,"citation_checks":[],"claim_id":52403,"claim_text":"The exo project claims that Apple featured its technology on new M5 Ultra Mac Studio and M6/M5 Pro Mac Mini pages, emphasizing low-latency RDMA over Thunderbolt 5.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-08-27T06:50:00.397459+00:00","source_name":"Latent Space","source_url":"https://www.latent.space/p/ainews-hot-chips-openais-jalapeno"},{"assertion_indexes":[4,19],"citation_authority_claim_id":42896,"citation_checks":[],"claim_id":42896,"claim_text":"Muse Spark 1.2 is available under two model IDs: muse-spark-1.2 priced at $1.25/million input tokens and $4.25/million output tokens, and muse-spark-1.2-contributor priced at $0.10/million input tokens and $0.20/million output tokens.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-08-06T06:27:21.027646+00:00","source_name":"Simon Willison's Weblog","source_url":"https://simonwillison.net/2026/Aug/5/muse-code-and-muse-spark-12/#atom-everything"},{"assertion_indexes":[4,14,19],"citation_authority_claim_id":67854,"citation_checks":[],"claim_id":67854,"claim_text":"Meta's Muse Spark costs $1.25 per million input tokens and $4.25 per million output tokens.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-09-26T06:52:24.666317+00:00","source_name":"Big Technology","source_url":"https://www.bigtechnology.com/p/what-i-got-wrong-about-metas-muse"},{"assertion_indexes":[5,14],"citation_authority_claim_id":79703,"citation_checks":[],"claim_id":79703,"claim_text":"Muse Spark Open uses a mixture-of-experts architecture with 816 billion total parameters and 42 billion active parameters per token.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-16T06:55:41.022571+00:00","source_name":"AWS Machine Learning Blog","source_url":"https://aws.amazon.com/blogs/machine-learning/metas-muse-spark-open-now-available-in-amazon-sagemaker-jumpstart/"},{"assertion_indexes":[5,14],"citation_authority_claim_id":79704,"citation_checks":[],"claim_id":79704,"claim_text":"Muse Spark Open supports a context window of 1 million tokens.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-16T06:55:41.022571+00:00","source_name":"AWS Machine Learning Blog","source_url":"https://aws.amazon.com/blogs/machine-learning/metas-muse-spark-open-now-available-in-amazon-sagemaker-jumpstart/"},{"assertion_indexes":[6],"citation_authority_claim_id":15124,"citation_checks":[],"claim_id":15124,"claim_text":"Professional services firms with high confidentiality requirements are increasingly purchasing Mac Minis to run local AI models.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-07-10T03:49:41.584436+00:00","source_name":"Nate Jones","source_url":"https://www.youtube.com/watch?v=RaAFquzj5B8"},{"assertion_indexes":[7,9],"citation_authority_claim_id":73152,"citation_checks":[],"claim_id":73152,"claim_text":"Reflection AI unveiled its first frontier open-weight AI model named Beam on October 5, 2026.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-06T07:14:31.844450+00:00","source_name":"TechCrunch AI","source_url":"https://techcrunch.com/2026/10/05/reflection-debuts-beam-a-open-weight-ai-model-to-rival-chinese-models-at-lower-compute-cost/"},{"assertion_indexes":[7,9],"citation_authority_claim_id":73153,"citation_checks":[],"claim_id":73153,"claim_text":"Reflection AI claims its Beam model matches the performance of leading Chinese open models on advanced reasoning benchmarks at significantly lower costs.","correction_role":null,"corroborates_eligible":true,"label":"opinion","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-06T07:14:31.844450+00:00","source_name":"TechCrunch AI","source_url":"https://techcrunch.com/2026/10/05/reflection-debuts-beam-a-open-weight-ai-model-to-rival-chinese-models-at-lower-compute-cost/"},{"assertion_indexes":[7,9],"citation_authority_claim_id":73154,"citation_checks":[],"claim_id":73154,"claim_text":"Beam is a text-only mixture-of-experts model with 501 billion total parameters and 23 billion active parameters.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-06T07:14:31.844450+00:00","source_name":"TechCrunch AI","source_url":"https://techcrunch.com/2026/10/05/reflection-debuts-beam-a-open-weight-ai-model-to-rival-chinese-models-at-lower-compute-cost/"},{"assertion_indexes":[7,9],"citation_authority_claim_id":73155,"citation_checks":[],"claim_id":73155,"claim_text":"Beam was pretrained on 23.8 trillion tokens and supports a 1 million token context window.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-06T07:14:31.844450+00:00","source_name":"TechCrunch AI","source_url":"https://techcrunch.com/2026/10/05/reflection-debuts-beam-a-open-weight-ai-model-to-rival-chinese-models-at-lower-compute-cost/"},{"assertion_indexes":[7,9],"citation_authority_claim_id":73158,"citation_checks":[],"claim_id":73158,"claim_text":"Reflection AI positions Beam as a direct rival to closed labs like Anthropic and OpenAI, Chinese open models, and Western players like Mistral, Meta, and Cohere.","correction_role":null,"corroborates_eligible":true,"label":"opinion","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-06T07:14:31.844450+00:00","source_name":"TechCrunch AI","source_url":"https://techcrunch.com/2026/10/05/reflection-debuts-beam-a-open-weight-ai-model-to-rival-chinese-models-at-lower-compute-cost/"},{"assertion_indexes":[7,8],"citation_authority_claim_id":80333,"citation_checks":[],"claim_id":80333,"claim_text":"Reflection released an open-source 501-billion-parameter Mixture-of-Experts language model named Beam.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:43:29.396966+00:00","source_name":"Hugging Face Blog","source_url":"https://huggingface.co/blog/reflection-beam"},{"assertion_indexes":[7,8],"citation_authority_claim_id":80334,"citation_checks":[],"claim_id":80334,"claim_text":"Beam contains 23 billion active parameters per token, with 256 experts per layer and 8 routed per token.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:43:29.396966+00:00","source_name":"Hugging Face Blog","source_url":"https://huggingface.co/blog/reflection-beam"},{"assertion_indexes":[7,8],"citation_authority_claim_id":80335,"citation_checks":[],"claim_id":80335,"claim_text":"Beam supports a context window of 1 million tokens using a hybrid attention mechanism with 43 sliding-window layers and 9 global layers.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:43:29.396966+00:00","source_name":"Hugging Face Blog","source_url":"https://huggingface.co/blog/reflection-beam"},{"assertion_indexes":[7,16],"citation_authority_claim_id":80388,"citation_checks":[],"claim_id":80388,"claim_text":"The Hugging Face Transformers library version 5.20.1 includes a new model named Beam from Reflection AI.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:45:12.374410+00:00","source_name":"Transformers Releases","source_url":"https://github.com/huggingface/transformers/releases/tag/v5.20.1"},{"assertion_indexes":[7,16],"citation_authority_claim_id":80391,"citation_checks":[],"claim_id":80391,"claim_text":"The Transformers library version 5.20.1 fixes a bug in auto mappings.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:45:12.374410+00:00","source_name":"Transformers Releases","source_url":"https://github.com/huggingface/transformers/releases/tag/v5.20.1"},{"assertion_indexes":[7,16],"citation_authority_claim_id":80392,"citation_checks":[],"claim_id":80392,"claim_text":"The Transformers library version 5.20.0 broke loading models on older torch versions.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:45:12.374410+00:00","source_name":"Transformers Releases","source_url":"https://github.com/huggingface/transformers/releases/tag/v5.20.1"},{"assertion_indexes":[9,12],"citation_authority_claim_id":76772,"citation_checks":[],"claim_id":76772,"claim_text":"Reflection announced Beam, a 501 billion parameter MoE LLM, which is behind Chinese models like GLM-5.3 in benchmark scores but claims higher token efficiency.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-13T06:44:17.954737+00:00","source_name":"Import AI","source_url":"https://importai.substack.com/p/import-ai-476-creativity-and-rsi"},{"assertion_indexes":[10,13,20],"citation_authority_claim_id":80512,"citation_checks":[],"claim_id":80512,"claim_text":"Google Research is integrating AI agents with physical wet-lab robotics to enable closed-loop molecular design cycles for discovering protein binders with single-digit nanomolar affinity for cancer targets.","correction_role":null,"corroborates_eligible":true,"label":"forecast","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:49:18.889096+00:00","source_name":"Google Research Blog","source_url":"https://research.google/blog/grounding-ai-agents-in-biological-reality-building-blocks-for-autonomous-wet-labs/"},{"assertion_indexes":[10,13,20],"citation_authority_claim_id":80513,"citation_checks":[],"claim_id":80513,"claim_text":"The VERID (VERified Infrastructure for Discovery Laboratories) framework is a semantic digital twin built on the open-source PyLabRobot that provides live 3D state tracking, virtual dry runs, hardware safety checks, and synchronized video logs to bridge the gap between AI reasoning and physical lab operations.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:49:18.889096+00:00","source_name":"Google Research Blog","source_url":"https://research.google/blog/grounding-ai-agents-in-biological-reality-building-blocks-for-autonomous-wet-labs/"},{"assertion_indexes":[10,13,20],"citation_authority_claim_id":80514,"citation_checks":[],"claim_id":80514,"claim_text":"In aptamer discovery experiments targeting myoglobin and \u03b1-thrombin, AI-driven design cycles raised confirmed on-target hit rates to 33.5% and 42.1% respectively, after agents autonomously identified and corrected wet-lab failure modes such as primer overhang disruption and non-specific binding.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:49:18.889096+00:00","source_name":"Google Research Blog","source_url":"https://research.google/blog/grounding-ai-agents-in-biological-reality-building-blocks-for-autonomous-wet-labs/"},{"assertion_indexes":[11],"citation_authority_claim_id":16112,"citation_checks":[],"claim_id":16112,"claim_text":"An AI co-scientist system generated drug targets for liver fibrosis that were subsequently proven efficient in wet-lab experiments by experts.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-07-11T03:00:40.889134+00:00","source_name":"AI Engineer","source_url":"https://www.youtube.com/watch?v=wNH3q9pqn0U"},{"assertion_indexes":[15],"citation_authority_claim_id":73418,"citation_checks":[],"claim_id":73418,"claim_text":"Black Forest Labs launched FLUX 3 Image, a model capable of native generation up to 4K, up to ten reference images, bounding-box layout control, and targeted multi-turn editing.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-07T06:43:30.759186+00:00","source_name":"Latent Space","source_url":"https://www.latent.space/p/ainews-pi-10-pi-durable-and-aie-nyc"},{"assertion_indexes":[15],"citation_authority_claim_id":73420,"citation_checks":[],"claim_id":73420,"claim_text":"Black Forest Labs announced commercial weights for FLUX 3 Image are available, with an open-weight variant promised in coming weeks, and hosted access through fal and Krea.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-07T06:43:30.759186+00:00","source_name":"Latent Space","source_url":"https://www.latent.space/p/ainews-pi-10-pi-durable-and-aie-nyc"},{"assertion_indexes":[15],"citation_authority_claim_id":80117,"citation_checks":[],"claim_id":80117,"claim_text":"Black Forest Labs released FLUX 3 Image Klein, a 5.6B parameter open-weight model for text-to-image generation, image editing, and multi-reference generation.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:36:12.754658+00:00","source_name":"Black Forest Labs","source_url":"https://bfl.ai/blog/flux-3-image-klein"},{"assertion_indexes":[15],"citation_authority_claim_id":80118,"citation_checks":[],"claim_id":80118,"claim_text":"FLUX 3 Image Klein shares core capabilities with the full FLUX 3 Image model, including native 4K generation, pixel-perfect editing, and bounding box controls.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:36:12.754658+00:00","source_name":"Black Forest Labs","source_url":"https://bfl.ai/blog/flux-3-image-klein"},{"assertion_indexes":[15],"citation_authority_claim_id":80119,"citation_checks":[],"claim_id":80119,"claim_text":"FLUX 3 Image Klein builds on FLUX 3's shared learnings across images, video, and audio, using spatial detail from images, motion from video, and event information from audio.","correction_role":null,"corroborates_eligible":true,"label":"fact","paired_claim_ids":[],"provenance_tier":"primary","recorded_at":"2026-10-17T06:36:12.754658+00:00","source_name":"Black Forest Labs","source_url":"https://bfl.ai/blog/flux-3-image-klein"}],"version":"evidence-bundle-v2"}